Letter to the Editor Regarding “Ticagrelor for Asian Patients With Acute Coronary Syndrome in Real-World Practice: A Systematic Review and Meta-analysis of Observational Studies.” The Benefit of Using Person-Year When the Follow-up Period Varies in a Meta-analysis

Abstract

Sir, We read with great interest a meta-analysis published in the Indian Heart Journal “Ticagrelor for Asian patients with acute coronary syndrome in real-world practice: a systematic review and meta-analysis of observational studies”. 1 We would like to congratulate the authors for their study on the effectiveness of ticagrelor vs clopidogrel in patients with acute myocardial infarction, by considering the data from published observational studies. The study has definitely provided us with better evidence in this field. However, we had the following observations to make: 1. Of the six studies included by the authors in the meta-analysis, each study had a different follow-up period. Also, there is a variation in the follow-up period for different cohorts (ticagrelor vs clopidogrel) within the study. An example for this discrepancy is a study by Lee et al included in the meta-analysis, 2 in which clopidogrel cohort was recruited and followed up from 2012 to 2014, while the ticagrelor cohort was recruited and followed up from 2013 to 2014. This implies that the first patient recruited in the clopidogrel cohort was followed up for three years unless he had a event of interest, and the first recruit of the ticagrelor cohort was followed up for a total period of two years. By logic, incidence rate of events increases with increase in the duration of follow-up. To avoid such discrepancies, a better method is to use patient-year in the denominator, which is calculated by multiplying the number of cohort recruits with the year of follow-up. 3 The pooled odds ratio (OR) of major bleeding for ticagrelor vs clopidogrel considering patient-year in the denominator is 1.40 (95% confidence interval [CI]: 0.92–2.13, I2 = 51%, p = 0.11) (Fig. 1) as compared with the OR for major bleeding reported by authors of meta-analysis (OR = 1.11 [95% CI: 0.62–2.00, I2 = 75%, p = 0.73]. Fig. 1 Pooled odds ratio and 95% confidence intervals (CIs) for major bleeding (total cohort population expressed as person-year) using a random-effects model by the method of inverse variance and heterogeneity assessed with the I2 measure. Fig. 1 2. Though as a rule of thumb, test for publication bias is to be performed when the number of included studies is more than ten as the power of tests otherwise is too low to distinguish chance from real asymmetry, inclusion of funnel plot in the meta-analysis would help the reader in assessing for any publication or other biases with respect to each outcome. 4 (Fig. 2) Fig. 2 Funnel plot (major bleeding). Fig. 2 Source(s) of support Nil. Presentation at a meeting Nil. Conflict of interest None.